Introduction

Substantial numbers of people experience at least one traumatic event during their lifetimes [1]. Although a traumatic life event could be a risk factor for the precipitation of various psychiatric disorders, most people can adapt to such an event with little disruption or recover their baseline level of functioning after a transient symptomatic period [2, 3]. Based on such observations, researchers become increasingly interested in determining which factors mediate individual differences in responding to adversity. The term resilience refers to the positive side of individual differences in people’s responses to stress and adversity [4]. Most early researchers studied it in children who successfully grew up in high-risk environment. Since then, researchers have applied the concept of resilience more broadly to individuals experiencing various traumatic life events, including physical illnesses, such as cancer [5, 6] and mobility disabilities [7] including spinal cord injuries [8], and psychiatric illnesses including posttraumatic stress disorder (PTSD) [9]. Indeed, resilience is important in maintaining quality of life [6, 7], emotional well-being [5], and functional independence [8] in spite of suffering such illnesses.

Although resilience had no clearly established definition, various contributing personal and environmental factors that might seem to work together were found [10]. Researchers have suggested that demographic variables, such as male gender, greater age, and higher education, personal attributes, such as internal locus of control and active coping strategies [11], positive psychological factors, such as hope, optimism, gratitude, and purpose in life [12], and socio-contextual factors, such as supportive relationships and community resources including family cohesion, friendship, and religious activities [4, 13], contribute to resilience in both children and adults. In addition, since resilience is a dynamic concept, interacting with adversity, various traumatic experiences should be considered together. For example, history of childhood maltreatment was reported to predict low resilience in community samples [14], while life-threatening diseases result in posttraumatic growth in some patients [15]. Of the suggested resilience factors, psychological factors, including psychiatric symptoms and positive psychological factors, might have clinical value, because they are potentially modifiable.

Among psychiatric illnesses, substantial research and reviews on resilience have focused on PTSD to date. However, PTSD is not the only pathological flip side of resilience. Indeed, a quantity of epidemiological and biological data has shown that traumatic life events, in either childhood or adulthood, may correlate broadly with depression [16], anxiety disorders [17], psychosomatic disorders [18], substance-related disorders, and antisocial behavior [19]. Therefore, research on resilience in these various trauma-related disorders is needed. Among these disorders, depression and anxiety disorders have clinical importance due to their high prevalence, accompanying substantial disability, and high recurrence rates [20]. Moreover, resilience was reported to mediate reduced depression and anxiety in otherwise healthy individual [21, 22], and researchers have proposed that patients with depression and/or anxiety disorders might receive the greatest benefits from resilience-enhancing interventions [23, 24].

When considering implications of resilience for psychiatric illnesses, researchers regarded resilience as protecting individuals against the development of an illness after adversity at first. Individuals with high resilience had a lower risk of having PTSD after traumatic life events [10, 25]. However, resilience might not completely protect from psychopathology [26]. Given that resilience allow individuals to cope well with traumatic events and to maintain relatively stable levels of functioning and quality of life, resilience in patients who developed psychopathology may be important in management and recovery from their illnesses. In agreement with this hypothesis, resilience has been proposed to have a prognostic value based on the finding that high resilience correlated with the favorable response to treatment in patients with PTSD [27]. In addition, certain studies reported resilience level changed following treatment in patients with depression [28] and with PTSD [27]. Taken together, these studies implied that resilience might influence patient’s prognosis, and some aspects of resilience could be enhanced by intervention. However, resilience might be difficult to address in clinical practice because of its complexity. The exact natures of the meaningful factors contributing to resilience in such patient have yet to be elucidated.

Given this background, the present study aimed to investigate characteristics of patients with depression and/or anxiety disorders who showed low resilience. We focused on psychological factors including psychiatric symptoms and positive psychological factors because of their potential modifiability. Researchers have suggested that individual vulnerability and resilience factors, as well as their psychopathologies, should be involved in future diagnostic system for personalized diagnosis and care [29]. Our results could provide a basis for recognizing clinically useful resilience factors and designing resilience-focused interventions in patients with depression and anxiety disorders.

Methods

Participants

During the 12-month study period between May 2009 and April 2010, patients who firstly visited Anxiety and Mood disorder Clinic at Seoul St. Mary’s Hospital, The Catholic University of Korea, and met the DSM-IV criteria for depressive and/or anxiety disorders were recruited consecutively. Diagnosis was conducted by a psychiatrist using semi-structured interviews of the Mini-International Neuropsychiatric Interview (M.I.N.I.) [30]. Eligibility criteria were being 18–65 years of age and literate in Korean. Exclusion criteria included a lifetime diagnosis of psychotic disorder, bipolar disorder, mental retardation, and any mental disorder due to general medical condition [25]. We also excluded individuals with significant personality disorders and/or medical problems likely to interfere with their study participation. A total of 125 psychiatric outpatients who met the inclusion and exclusion criteria consented to participate in this study. Restricting analyses to examining data from who had completed all measures, the final sample included 121 (of 125) patients (mean age, 36.0 ± 13.5; 51.2 % were female). The study procedure was approved by the Institutional Review Boards of the ethical committee of the Seoul St. Mary’s Hospital at the Catholic University of Korea.

Demographic, clinical, and psychological measures

During clinical interviews, we assessed patients for the demographic variables (such as education years, marital status, employment status, monthly income, religion, and physical exercise) and clinical characteristics (including medical illness and psychiatric family history) that researchers have suggested may influence resilience [1, 4, 13].

Both childhood maltreatment and other types of trauma experiences were assessed as trauma load because they would influence resilience levels [1, 31]. Parent–child conflict tactics scales (PCCTS) [32] were employed to measure the types and frequencies of childhood maltreatment. Five psychological aggression items, nine physical assault items, and two supplemental items about sexual maltreatment of PCCTS evaluate each participant’s experiences regarding maltreatment before the age of 18. Additionally, other potentially traumatic events during the lifetime were measured with the life events checklist (LEC) [33]. The LEC inquires about multiple degrees of exposure to each trauma, using a 5-point nominal scale. Among various degrees of exposure, a score 1 (happened to me) was regarded as an “experienced” trauma with exception that a score of 2 (witnessed it) was regarded as “experienced” in items 14 and 15 in a previous study [34]. Score of LEC was then calculated by summing the numbers of experienced events. The PCCTS [35] and LEC [36] have been translated into Korean and adapted for use in the Korean population.

To measure resilience, the Korean version of Connor–Davidson resilience scale (CD–RISC) was used [37]. The CD–RISC was developed for clinical practice as a measurement of coping ability in the face of adversity [38]. It consists of 25 items, and each item is rated on a 5-point Likert scale, ranging from 0 (not true at all) to 4 (true nearly all the time). Higher total scores indicate greater resilience. The CD–RISC is regarded as one of the resilience measures having the best psychometric properties in a meta-analysis [39]. In addition, it is able to evaluate changes of resilience in response to interventions [40]. We used the total CD–RISC scores due to instability of the 5-factor structure.

We measured positive psychological factors comprising optimism, gratitude, humor, hope, spirituality, and purpose in life using the following self-report questionnaires: the life orientation test-revised (LOT-R) [41], the gratitude questionnaire (GQ-6) [42], the sense of humor questionnaire (SHQ-6) [43], the state hope scale (SHS) [44], the functional assessment of chronic illness therapy–spirituality (FACIT–Sp) [45], and the “purpose in life” test (PIL) [46], respectively. LOT-R [47], SHQ-6 [48], SHS [49], and PIL [50] have been translated and adapted for use in Korean population, and GQ-6 was validated in Korean population [51]. The Korean version of FACIT–Sp was licensed from http://FACIT.org.

Among psychiatric symptoms, we assessed participants’ symptoms of depression, anxiety, somatization, hostility, problematic alcohol use using the Beck depression inventory (BDI) [52], the state-trait anxiety inventory (STAI) [53], somatization and hostility subscales of the symptom checklist 90-revised (SCL-90-R) [54], and the alcohol use disorder identification test (AUDIT) [55], respectively, based on their associations with traumatic life events as mentioned above. Korean versions of BDI [56], STAI [57], SCL-90-R [58], and AUDIT [59] were validated.

Data analysis

To identify characteristics of patients having different resilience levels, we divided patients into 3 categories based on their CD–RISC score percentile, as was done in a previous study [60]. We defined the high-resilience group as having CD–RISC scores ≥75th percentile, the medium-resilience group as having scores ≥25th percentile and <75th percentile, and the low-resilience group as having scores <25th percentile. To compare these three groups for continuous and categorical variables, we performed one-way analyses of variance (ANOVA) and χ2 tests, respectively. Post hoc multiple comparisons were performed by Bonferroni adjustment.

Next, variables associated with the levels of resilience in the univariate analysis (P < 0.1) were entered into the multinomial logistic regression models. After designating the low- and medium-resilience group as the outcome and the high-resilience group as the reference, a series of multinomial logistic regression were performed to examine which variables independently predicted lower resilience (i.e., the low- or medium-resilience groups). Dividing variables into the three categories of demographic and trauma loads, psychiatric symptoms, and positive psychological factors, we attempted to identify the significant predictors of each categories, and then, we attempted to find the most significant predictors of all the variables. Since substantial intercorrelations between psychiatric symptoms and positive psychological factors and possible multicollinearity were expected, forward selection method was used for the regression models including these factors. In the first step, relevant factors including demographic characteristics and traumatic experiences were simultaneously entered into the model. In the second step, psychiatric symptoms and positive psychological factors were entered in separate models using forward selection method after adjusting for the demographic and clinical characteristics (partially adjusted model). In the final model, all psychological variables were entered into a model using forward selection method after adjustment of demographic and trauma load covariates (fully adjusted model). To perform the assumption checking for each model’s goodness of fit, we used the likelihood ratio test. Statistical significance was set at P < 0.05, two tailed.

Results

The 121 participating patients had the following principal psychiatric diagnoses: 80 (66.1 %) patients had depressive disorders comprising major depressive disorder (N = 65), dysthymic disorder (N = 6), and depressive disorder not otherwise specified (NOS) (N = 9). Remaining 41 (33.9 %) patients had anxiety disorders comprising panic disorder (N = 26), generalized anxiety disorder (N = 10), obsessive compulsive disorder (N = 9), social anxiety disorder (N = 9), PTSD (N = 1), and anxiety disorder NOS (N = 5). Among them, 20 (16.5 %) patients were diagnosed as having both depressive and anxiety disorders. Mean (±SD) scores on the CD–RISC score was 48.5 (±19.8) in all patients. Although CD–RISC score did not significantly differ according to their principal diagnosis (P = 0.059), patients with depressive disorders tended to have lower CD–RISC score (46.1 ± 18.7) than those with anxiety disorders (53.6 ± 21.0).

Characteristics of the low-, medium-, and high-resilience groups

Table 1 summarizes the demographic and clinical characteristics of the low-, medium-, and high-resilience groups. Mean age, exercise frequency, and degree of childhood emotional maltreatment differed significantly across the three groups. Post hoc analysis revealed that the low-resilience group had younger age and less frequent exercise than the high-resilience group had (P = 0.012 and P = 0.048, respectively). The low-resilience group recalled more childhood maltreatment in the form of emotional aggression than did either the medium- or the high-resilience groups (P = 0.003 and P = 0.001, respectively). The patients’ psychiatric symptom scores (except for problematic alcohol use) and scores on all positive psychological factors we considered increased or decreased, respectively, in the following order: the low-, the medium-, and the high-resilience groups.

Table 1 Demographic and clinical characteristics of the low-, medium-, and high-resilience groups in patients with depressive and anxiety disorders
Full size table

Predictors of the low- and medium-resilience groups versus high-resilience group

Table 2 summarizes the results of the multinomial logistic regression analyses examining predictors of the low- and medium-resilience groups with the high-resilience group as the reference. Among demographic and trauma loads, younger age (P = 0.039) and more numbers of lifetime trauma experiences (P = 0.049) were significantly associated with the medium- versus the high-resilience group (model 1). Among psychiatric symptoms, more severe trait anxiety predicted both the low (P < 0.001)- and medium-resilience groups (P = 0.002) versus the high-resilience group after controlling for demographic and trauma loads (model 2). In positive psychological factors, lower spirituality (P = 0.005) and purpose in life (P = 0.026) characterized the low- versus high-resilience group, whereas low sense of humor (P = 0.035) characterized the medium- versus high-resilience group after controlling for demographic and trauma loads (model 3). In the final model including all variables using the forward stepwise procedure (model 4), spirituality was found to be a key factor predicting both the low (P = 0.001)- and medium-resilience groups (P = 0.029) versus high-resilience group. In addition, lower purpose in life (P = 0.021) and less frequent exercise (P = 0.043) were significantly associated with the low- and medium-resilience group, respectively. The likelihood ratio test revealed that the model fits were statistically significant in every model.

Table 2 Predictors of the low- and medium-resilience group versus the high-resilience group in patients with depression and anxiety disorders
Full size table

Discussion

Although resilience has been extensively studied in developmental perspectives and among healthy individuals, relatively little is known which factors are associated with resilience in patients with psychiatric illnesses [2]. To our knowledge, this study is the first attempt to examine clinical and psychosocial factors associated with different levels of resilience in patients with depression and/or anxiety disorders. We found that low spirituality and purpose in life predicted low- versus medium-resilience group, whereas low spirituality and less frequent physical exercise predicted medium- versus high-resilience group in a sample of outpatients with depression and/or anxiety disorders. Among psychiatric symptoms, trait anxiety characterized low- and medium-resilience group after controlling demographic and trauma factors although it was not included the final model. These factors might help in predicting successful adaptation in response to depression and anxiety disorders. Since depression and anxiety disorders frequently fail to remit, recur easily, and persist [20], such patients might need additional clinical interventions to improve overall prognosis. In this regard, our findings could provide data for use in the development and application of resilience-enhancement interventions in patients with depression and/or anxiety disorders.

While the significance of the majority of positive psychological factors are well established for the general population [61, 62], relatively little is known about their clinical significance in patients with psychiatric illnesses. Most notably in this study, low spirituality was the key independent predictor associated with low resilience in patients with depression and/or anxiety disorders. Spirituality is defined as “the way in which people understand their lives in view of their ultimate meaning and value” [63]. It involves a sense of meaning and purpose, as well as peace and harmony, and stands distinct from religiosity [45]. Given that the purpose in life signifies the ability to find positive meaning in an adverse event [64], spirituality and purpose in life seem to show an overlap. Accordingly, purpose in life also independently predicted the low-resilience group in this study. Based on the definitions of spirituality and purpose in life, one can speculate that cognitive process of meaning finding in the face of adversity as well as emotion regulation ability to maintain peace and harmony may be helpful in keeping relatively high resilience despite their psychiatric illnesses. Our findings well corresponded with previous report in which a sense of meaning and purpose in life contributed to resilience, recovery, and posttraumatic growth after various traumatic life events [25, 65].

Clinical implications of spirituality and purpose in life were also suggested in relation to psychiatric and physical health problems. For instance, the strong inverse relation between spirituality and depression has been reported [66, 67]. Spiritual well-being associates negatively with suicidal ideation in terminally ill cancer patients [68], whereas purpose in life has been linked to better physical and mental health outcomes [69] and even longevity [70]. Moreover, spirituality has been proposed to be related to serotonin system [71], and purpose in life has been associated with cardiovascular risk factors and immune markers [72]. These results proposed that these two positive psychological factors would be meaningful for applying to various clinical populations. In this context, our findings emphasize the significance of spirituality and purpose in life in enhancing resilience of patients with depression and/or anxiety disorders. Spirituality or meaning-focused intervention, such as logotherapy [64] or meaning-making intervention [73], might be valuable psychosocial interventions in patients with relatively lower resilience.

Of various psychiatric symptoms, severe trait anxiety was the most reliable predictor for lower resilience in patients with depression and/or anxiety disorders. Although trait anxiety was not remained in the final model and this may imply the significance of positive psychological factors over psychiatric symptoms, trait anxiety as the most influencing psychiatric symptom on resilience is still worth mentioning in patients with depression and/or anxiety disorders. Although the majority of previous studies reported a significant negative correlation between resilience and depressive symptoms [74, 75], we found trait anxiety has greater influence on resilience prediction than depressive symptoms have. This finding is consistent with previous report of strong negative relationship between neuroticism and resilience in healthy young adults [76] and the association between high trait anxieties with vulnerability to depression when facing adversity [77].

Trait anxiety refers to general and long-standing feelings of apprehension, tension, nervousness, and worry [53]. Biologically, it occurs in association with amygdala reactivity to threat-related cues [78]. Considering the roles of amygdala in fear conditioning, reconsolidation, and extinction after stressful life events, the relationship between trait anxiety and resilience agrees with the idea that a capacity to avoid overgeneralizing specific stimuli and to facilitate extinction may characterize resilience [79]. Fear circuit centered on amygdala may serve as a convergent neural correlate between trait anxiety and resilience. Our finding addressing trait anxiety has an important implication, in that clinicians would propose the need of evaluation and management of trait anxiety for enhancing patient’s resilience in addition to using positive psychological approaches. In spite of the name “trait anxiety”, a recent study showed that environmental factors, such as perceived social support, can mitigate trait-like anxiety in healthy subjects [80]. Furthermore, pharmacological interventions, such as benzodiazepine partial agonist, corticosterone, and selective neurokinin-1 receptor agonist, also attenuated enhanced fear response in a mouse model of trait anxiety [81]. Future research on effective pharmacological and psychosocial interventions for trait anxiety is encouraged.

Apart from spirituality and trait anxiety, some possible influencing factors were only significant predicting medium- versus high-resilience group but not in predicting low- versus high-resilience group. Among them, the importance of frequent physical exercise was noted. Physical exercise, especially aerobic exercise training, has been proposed as protective against sensitivity to stress, depression, and anxiety in both clinical and general population [22, 82]. Our finding would proposed that recommending physical exercise would be helpful in enhancing resilience among patients with depression and/or anxiety disorders, especially those with having medium levels of resilience. Additionally, the relationships between greater age, fewer traumatic experiences, and sense of humor with higher resilience were observed, and these were also suggested in previous studies among general populations [1, 14, 16]. However, these factors were only significant in predicting medium- versus high-resilience group and did not remain to be significant in the final model adjusting for all other factors in the sample of patients with depression and/or anxiety disorders. These results differentiating predictors of the low- from medium-resilience group support the notion that meaningful resilience factors could be different according to the characteristics of sample, such as general population or clinical samples. Future replication studies in patients with depression and/or anxiety disorders will be needed to confirm our findings.

In addition, resilience levels of our participants who were outpatients with depression and/or anxiety disorders are needed to be discussed with respect to previous reports. The mean CD–RISC score of 48.7 ± 19.8 in this study is lower than that of Korean general population including nurse, university students, and firefighter (61.2 ± 13.0) [37]. Lower resilience in psychiatric outpatients was consistent with the report in US population [38] whereby CD–RISC score of general population (80.4 ± 12.8) was significantly higher than that of primary care patients (71.8 ± 18.4) and that of primary care patients was higher than that of psychiatric outpatients (68.0 ± 15.3), patients with generalized anxiety disorder (47.8 ± 19.5), and PTSD patients (52.8 ± 20.4). The fact that patients with high-resilience group (74.9 ± 11.6) showed comparable levels of resilience to general population is worthy of notice. This may reflect that resilience would be an independent clinical factor from psychiatric illnesses and symptoms. Interestingly, Koreans showed relatively lower-resilience scores both in general population and psychiatric outpatients. Alike, Chinese adolescents experienced earthquake and Turkish earthquake survivors showed the mean CD–RISC score of 64.9 ± 13.3 [83] and 70.1 ± 14.1 [84], respectively. The fact that Asians were reported to tend to select midpoint on items involving positive emotion than Americans [85] may be one of the possible explanations for the differences in levels of resilience according to the countries. Therefore, levels of resilience might be interpreted in considering the demographic, clinical, and ethnic differences [37].

Several limitations are needed to be addressed in the present study. First, we considered patients with depression and/or anxiety disorders as a whole, although difference in resilience levels according to diagnosis was suggested in a study of Connor and Davidson [38]. However, some researchers have proposed depression and anxiety disorders as just different expressions of an emotional disorder, sharing general vulnerability factors, based on their high rates of comorbidity, and symptomatic overlaps in clinical settings [86]. Moreover, common psychotherapeutic approaches, such as the unified protocol, have also been developed as effective intervention strategies for broad-spectrum patients with depression and anxiety disorders [87, 88]. Second, clinical features of participants, such as outpatient population of a major university hospital and 16.5 % of comorbidity of depression and anxiety disorders, and Korean ethnicity may limit the generalization of our findings. Third, factors influencing resilience, among psychological factors in particular, were correlated with each other to some degrees. For example, strong positive correlations between trait anxiety, state anxiety, and depression were found. Therefore, we used logistic regression analysis with forward selection method in analyzing psychological factors after adjusting demographic and trauma covariates. More distinctive and comprehensive classifications of resilience factors with the involvement cognitive strategies and environmental factors will be desired in future studies. Lastly, an inherent limitation of this study is the inability to establish causality because of the cross-sectional design. Although resilience may be a physiological or psychological process rather than a static characteristic [89], resilience was measured by self-reported questionnaire at certain point. Longitudinal studies are needed to confirm the roles of observed predictors for resilience and the relationship between resilience and better coping in patients with depression and/or anxiety disorders.

Conclusion

Depression and/or anxiety disorders could be regarded as adversity to be dealt with, and thus, resilience factors might be meaningful in coping and recovery processes. We found that spirituality, purpose in life, trait anxiety, and physical exercise might contribute to different levels of resilience in patients with depression and/or anxiety disorders. These results emphaizes the value of positive psychological factors, spirituality, and purpose in life in particular, as well as the possible importance of trait anxiety for predicting and for enhancing resilience in such patients. Although more clinical implications of resilience in these patients need to be elucidated, resilience seems to be modifiable [9, 28] and it has been regarded to be associated with better prognosis and coping with illnesses [9]. In this context, our results might deepen the understanding of resilience in patients with depression and/or anxiety disorders and propose the potential targets for resilience-enhancement intervention, such as enhancing spirituality and purpose in life, in patients with low to medium levels of resilience for better outcome.